Quality inspection of bakery products using a color-based machine vision system

Muffins were evaluated for color by visual examination and by development o f a machine-reading system coupled with discriminant analysis of the data a cquired. A classification algorithm separated light from dark-colored muffi ns. The system's precision was assessed by evaluating the color of 4 cm dia meter muffins pregraded prior to the evaluation of color and without pregra ding. Applied to 200 samples, the automated system was able to correctly cl assify 96% of the pregraded and 79% of the ungraded muffins. The algorithm procedure was able to classify muffins at an accuracy level better than 88% in most cases whereas quality decisions among inspectors varied by 20 to 3 0%. Critical to precision by the machine-read procedure was control of the illumination.